The integration of GPS into cel-lular telephones enables a po-tentially vast array of new appli-cations ranging from consumer

gimmicks through efficiency multipliers forenterprises to lifesaving safety and securityapplications. In the United States, the En-hanced-911 regulations have been and re-main a major catalyst for this deployment.In Europe, the commercial potential of lo-cation-based services (LBS) is driving it. Re-gardless of the drivers, the technology con-vergence is happening with an ever-increasing momentum.

These new applications and the cell-phone environment itself, however, pose sig-nificant challenges for the GPS community.They demand GPS solutions that can be im-plemented in tiny spaces at extremely lowcost in extremely high volumes, operate re-liably in a much broader range of environ-ments than was hitherto considered possi-ble, acquire signals in seconds under extremeconditions and, for some applications, do sowithout the aid of stored orbital data.

These problems and the availability ofcellular communications itself spawned theconcept of assisted GPS (AGPS) in whichthe network assists the GPS receiver to per-form its various functions. This article re-ports on AGPS developments leveraging

cellular telephone networks to help acquireand deliver accurate GPS fixes from any-where, anytime.

System ConsiderationsIn developing a user system for AGPS, sev-eral factors must be considered, includingthe type of assistance to be provided by thenetwork, the type of cell-phone networkand the corresponding AGPS standards,and environmental factors such as radio fre-quency (RF) compatibility between theGPS module and the host platform.

Benefits of Assistance. There aremany types of assistance that can be pro-vided by the network to the GPS receiver.The receiver could be a fully functional re-ceiver capable of selecting satellites, acquir-ing signals, achieving time synchronization,extracting data, performing measurements,and computing its own navigation solution.Nevertheless, its acquisition speed can beenhanced through the provision of assis-tance. Furthermore, it can avoid spend-ing time to extract all of the required datafrom the satellite signals if most of this issupplied by the network.

Signal Levels. More importantly, therange of signal levels at which the receivercan operate can be greatly increased if thereceiver is relieved of the requirement to

extract the 50 bits-per-second navigationdata stream that is modulated onto the sig-nals. This data cannot be extracted in atimely manner (or at all in many cases) ifthe carrier-to-noise density ratio (C/N0) isbelow about –172 dBW (–142 dBm) butcode-phase measurements can still bemade for much weaker signals than this.Since the navigation data is not locationspecific, it can be supplied by a remote re-ceiver that has a clear line of sight to thesame satellites.

The data supplied by the network in thisway can include ephemeris coefficients, al-manac coefficients, satellite health data,satellite clock error coefficients, atmos-pheric error coefficients, and so on. Addi-tionally, excerpts from the data sequencecan be supplied to facilitate coherent in-tegration over periods much longer than anavigation data bit interval.

MS/UE Assisted. In principle, the re-ceiver can be simplified if some of the com-putation, such as the navigation solution,is performed by the network. This form ofoperation is referred to as mobilestation–assisted (MS-Assisted) or userequipment–assisted (UE-Assisted) in con-trast to solutions performed in the user’sequipment: MS-Based or UE-Based.

The advantage in terms of simplifyingthe handset is questionable given the com-putational capacity of modern cell-phoneand GPS chipsets. Furthermore, if the lo-cation information is required in the hand-set, this advantage is offset by the need tocommunicate back and forth. Neverthe-less, such architectures are commonplace.

The network has coarse informationabout the location of a receiver embed-ded in a cell phone. If it supplies this to thereceiver, along with its uncertainty, the re-ceiver then can use this to determine whichsatellites to search for, restrict its searchranges, and initialize its navigation solu-tion. The uncertainty typically is suppliedin the form of either the size and orienta-

AGPSINNOVATION ���

Assisted GPSUsing Cellular Telephone Networks for GPS Anywhere

40 GPS World MAY 2005 www.gpsworld.com

Rod Bryant, SigNav Pty. Ltd.

THE FIRST TWO ARTICLES TO APPEAR IN THE INNOVATION COLUMN IN GPS WORLD more than 15 yearsago were entitled “GPS: A Multipurpose System”and “The Limitations of GPS.” In the first article, GPSwas trumpeted as a revolutionary positioning and navigation technique that could be used in all sorts ofunexpected ways.This premise has stood the test of time with new uses for GPS still being discovered.The second article reminded readers that GPS may not be a panacea for all of our positioning needs— that there are some situations in which GPS fails us. In particular, it was noted that GPS signals are“blocked”by buildings, making indoor use of GPS impossible. But what was impossible 15 years ago ispossible today. New designs have greatly improved the sensitivity of GPS receivers so they can makecode-phase measurements even on the severely attenuated signals inside buildings. And if the signal istoo weak for the receiver to extract the satellite navigation message itself, the necessary data can besent to the receiver using a cellular telephone network, which also can supply timing information tohelp the GPS receiver acquire signals more quickly. In this month’s column, we will investigate how thisso-called “assisted GPS”works and why we can now say that GPS works (virtually) anywhere. — R.B.L.

www.gpsworld.com MAY 2005 GPS World 41

tion of the semi-major and semi-minor axesof an ellipsoid of uncertainty, or the hori-zontal radius and vertical height of a cylin-der of uncertainty.

Time Assistance. The network also cansupply time and its uncertainty. This may besupplied precisely using hardware in thehandset, or much more coarsely over the airfrom a server. If it is sufficiently precise (forexample, to within a few microseconds), itcan be used to restrict the search range forthe absolute code phases of the satellite sig-nals. If it is less precise (for example, only towithin many microseconds to a few seconds),it can be used to restrict the search range forthe relative code phases of the satellite sig-nals once an initial signal has been acquired.

If the time assistance is fairly precise (to afew milliseconds or better), the handset po-sition errors resulting from the uncertaintyin the satellite positions corresponding tothe time uncertainty will be small enough (afew meters) to be tolerated in most cases.

However, if it is coarser than this, the nav-igation solution also will have to solve forthe time error. Note that the receiver mayonly have code-phase measurements ratherthan full pseudorange measurements as usedin a normal receiver — requiring a proce-dure to resolve the one-millisecond code-phase ambiguities and a special algorithm tosolve for the time error.

Additional Assistance. Alternativelyto supplying position, time, and satellite-de-rived data, the network can derive and pro-vide other assistance including approximatecode phases at a certain instant in time, withthe corresponding uncertainties, as well asthe Doppler offsets with their uncertainties.

Assistance can be supplied in the “userplane” or in the “control plane.”

In the latter case, assistance is supplied viacommunication over the signaling channelsfrom a server integrated into the networkinfrastructure. The standards discussed in alater section relate to this form of assistance.

In the former case, the assistance is sup-plied from a user server (typically Web-based), using standard communicationschannels such as Short Messaging Service(SMS) and General Packet Radio Service(GPRS) over Global System for MobileCommunications (GSM) networks, or Sin-gle Carrier Radio Transmission Technology

(1xRTT) over code division multiple access(CDMA) networks.

CDMA vs. GSMThere are technical differences betweenGSM and CDMA cellular technologies thatimpact AGPS implementation in these net-works. Equally importantly, the two com-munities have evolved different AGPS stan-dards discussed in the next section.

Precise Timing. The first technical dif-ference relates to the fact that precise tim-ing is fundamental to CDMA operation. Thehandset synchronizes to the communica-tions code very precisely (that is, well belowthe microsecond level). Using hardware (forexample, a pulse and message), precise net-work time can be transferred to the GPS re-ceiver subsystem. Of course, this time willcontain an error equal to the communica-tion latency between the network and thehandset, but it is more than adequate as pre-cise time assistance for the AGPS purposesdescribed in the previous section.

GSM does not use spread spectrum codes,and hence this form of precise time assis-tance is not intrinsically available in a GSMnetwork. To deliver precise time assistance(with uncertainties of 5 or 10 microseconds),GSM networks have to be augmented. Notsurprisingly, few network operators are keento roll out additional infrastructure for thispurpose, and GSM deployment of AGPStypically is required to operate with coarsetime assistance.

Time Slots. Another technical differ-ence relates to the fact that GSM uses shorttime slots, so each handset communicates infrequent short bursts. CDMA handsets,on the other hand, communicate using muchlonger bursts. This impacts on the forms ofcell-phone interference mitigation tech-niques that can be employed by AGPS so-lutions in the two cases as will be discussedin a later section.

Cell Sizes. A third technical differencerelates to the fact that the GSM technologyhas a limitation on its cell sizes of around 35kilometers in radius. CDMA cell sizes, onthe other hand, are only limited by trans-mission power and relevant standards (suchas CDMA code-phase search ranges).Hence, they can be much larger. If the coarselocation assistance is derived from the cell

location, its uncertainty can be much largerin a CDMA system than in a GSM system.In practice, however, this is not a significantfactor because the most-demanding AGPSenvironments tend to be in inner-citieswhere cell sizes can be limited to a few kilo-meters in radius.

AGPS StandardsBoth the CDMA and the GSM communi-ties have developed standards for controlplane AGPS messaging (TIA/EIA/IS-801-1, 3GPP2 C.S0022-0-1, 3GPP TS 25.331)and for minimum operational performanceof AGPS handsets (TIA 916, 3GPP2C.P9004-0, 3GPP TS 25.171 V6.0.0).There is considerable similarity between theassistance fields included in the two proto-cols. The minimum performance standardsare measured in both cases using five sepa-rate statistical tests.

The five tests are of sensitivity, nominalaccuracy, dynamic range, multipath scenario,and moving scenario with periodic update.The nominal accuracy tests are for static ac-curacy under typical signal strength condi-tions rather than weak signal conditions andwith no multipath present. Performance inthe presence of multipath is tested separately,as are the performances under weak signalconditions and under typical land-based dy-namic conditions.

One difference between the two perform-ance standards is the handset must respondwithin 16 seconds in the CDMA case buthas 20 seconds to respond in the GSM case.In both cases the required sensitivity is –147dBm, and the horizontal positioning accu-racy, although defined differently, is simi-larly around 30 meters.

Another difference is the GSM standardallows for either precise or coarse time as-sistance. When only coarse time assistanceis provided, the sensitivity test is conductedwith one satellite at –142 dBm. This is arecognition there is a performance penaltyfor not providing precise time assistance.

The other main difference is that, in thecase of MS-Assisted (UE-Assisted) opera-tion, the CDMA standard calls for the ab-solute code-phase accuracy to be tested,whereas the GSM standard calls for the lo-cation to be computed in accordance with adefined algorithm and for the accuracy of

the result to be tested. This is more signifi-cant than it seems because the code-phasetest is of absolute code-phase accuracy ratherthan relative code-phase accuracy. To passthis test, the time assistance must be used todetermine the measurement instant withnanosecond precision.

The reason for this requirement is the lo-cation server can combine GPS and CDMAmeasurements in performing a hybrid fixonly if the absolute GPS code-phase meas-urements are known for a precise time (ac-cording to the CDMA handset’s local clock).

These standards have emerged from com-plex techno-political negotiations betweennetwork operators, handset manufacturers,technology providers, and semiconductormanufacturers. They represent negotiatedcompromises between these various groupsrather than a true consensus as to real-worldrequirements. In particular, the author con-siders the sensitivity requirement to be lack-ing in stringency. For reliable positioningunder most indoor conditions, sensitivity ofat least –150 dBm is essential and better than–153 dBm is desirable. Sensitivity of betterthan –185dBm is ideal.

Cell-Phone InterferenceOne of the technical problems facing theGPS cell-phone developer is the interfer-ence to GPS reception from the very strongcellular transmissions of the handset. Thisis an even more serious issue given that theGPS front-end and antenna performancetypically is compromised as a result of thesevere physical constraints on the design. Itis further exacerbated by the need for ex-treme sensitivity.

The ideal solution to this problem is toprovide sufficient filtering in the GPS RFpath to permit concurrent operation of thereceiver and handset transmitter. However,the significant benefits that flow from suchan approach come at a cost. In particular, themore complex RF design results in additionsto the bill of materials that add cost andspace.

Furthermore, it inevitably degrades thefront-end noise figure of the GPS receiver,thereby offsetting performance gains rela-tive to alternative approaches.

CDMA Mode Switching. In the case ofa CDMA handset, the main alternative is to

switch modes between GPS and the hand-set. The two subsystems cooperate so thereceiver “listens” only during timeslots whenthe handset is not permitted to transmit.

For example, during the 16-second re-sponse period, the transmitter may only beable to transmit for a fraction of the timein bursts of a few hundred milliseconds.While this process allows roaming to con-tinue, it may well represent an unacceptablelimitation on speech communication dur-ing that period. Meanwhile, the imposi-tion of antenna switching constraints onGPS signal processing also is significant asintegration periods must be sized to fit withinthese constraints.

GSM Signal Blanking. For GSM, theproblem is less extreme because the technol-ogy uses timeslots of only a few millisecondsin width. Signal blanking can be used dur-ing these slots, thereby efficiently avoidingthe need for cooperation between the soft-ware of the two subsystems.

Nevertheless, the blanking itself eats intothe GPS integration periods, thereby de-grading sensitivity and/or increasing typicalacquisition times.

Receiver ArchitecturesIn tailoring a GPS receiver for embeddedAGPS applications, several factors need tobe considered including correlator design,memory requirements, and microproces-sor control issues.

AGPS receivers need to acquire weak sig-nals quickly. To meet the demands of themarket and exceed the demands of theAGPS standards, more-sophisticated base-band hardware is needed than was requiredof conventional GPS receivers of only a fewyears ago. Essentially, this means many morecorrelator taps or “fingers.” How those fin-gers are organized and the type of signal pro-cessing employed are critical factors.

Limited Coherent Integration. Fig-ure 1 illustrates a common search engine

���INNOVATION AGPS

¶ Figure 1 A two-bin hardware search engine. The first set of mixers mix the final local oscillatorsignals with the incoming signal to downconvert the selected satellite signal to near baseband. Thefirst set of summing blocks perform coherent integration on the downconverted, despread signal. Thesquaring blocks compute the magnitudes of the complex integrals resulting from the coherent integra-tions. The second set of summing blocks perform non-coherent integration on the squarer outputs. Ashift register produces multiply delayed versions of the locally generated pseudorandom noise code.

42 GPS World MAY 2005 www.gpsworld.com

www.gpsworld.com MAY 2005 GPS World 43

architecture. This design performs multi-ple rounds of coherent integration for eachof the n fingers per frequency bin, and in-tegrates the results non-coherently. Thearchitecture lends itself to efficient hard-ware mechanization through reuse of thearithmetic elements. Chips have been pro-duced using this design to incorporate mul-tiple bins and 20,000 fingers or more. Otherdesigns, such as u-Nav Microelectronics’uN8130 baseband chip (with which the au-thor is very familiar), combines a 2,048 fin-ger 3 four-bin search engine with 12 four-finger correlators.

The first limitation of this hardwaresearch engine is the coherent integrationperiod is limited to much less than a navi-gation message bit. If not, the probabilityof bit transitions occurring within integra-tion periods will be high, and excessive ran-dom losses will result. The effect of thislimitation is to ensure the squaring lossesprior to the non-coherent integration arerelatively large. The end result is relativelylong overall integration periods are requiredto achieve the desired sensitivity.

The second limitation of this approachis that, when it is possible to constrain thesearch to a small range of code phases, therest of the fingers are effectively wasted.When precise time assistance is availablethis means most of the potential of thesearch engine hardware is wasted all of thetime. When only coarse time assistance isavailable, it means the full potential of thehardware is being utilized for acquiring thefirst satellite signal — but, again, most ofits capacity is wasted when acquiring sub-sequent satellite signals.

Flexible Design. Figure 2 illustrates analternative organization designed to addressthese issues. In this case, the signal process-ing of each bin is much more sophisticated,and hence the entire search engine nolonger lends itself to hardware mechaniza-tion. Instead, it is best implemented as a setof correlator channels each with multiplefingers. By using multiple channels to-gether, one or more larger search enginescan be built up as needed to span the re-quired code-phase search range.

This arrangement draws on the patentedsubATTO signal processing technology(see side bar), which facilitates coherent

integration over much longer intervals thana bit period. It results in shorter integrationperiods being used to achieve the same sen-sitivity. Since this more-flexible architec-ture also allows all of the hardware re-sources to be effective all of the time duringacquisition, it results in far more cost-ef-fective use of hardware to achieve the re-quired sensitivity and acquisition time.

This approach has been employed usingtwo very different hardware architectures.In the uN8130 baseband processor, a mod-estly dimensioned hardware search engineis used in tandem with 12 correlator chan-nels comprising four fingers each. The cor-relators are used for both acquisition andtracking while the search engine performsthe more difficult task in acquiring the firstsatellite and some of the subsequent satel-lites. Whenever the search engine acquiresa signal, it passes it to one of the correla-tors. This solution has been adapted suc-cessfully for both CDMA and GSMoperation, with per-formance well in ex-cess of the standards.

In another exam-ple, flexible correla-tor hardware re-sources have beenincorporated into achip designed to op-erate with a hostprocessor. subATTOprocessing takesplace on the host. Asearch strategy wasdevised that keeps thehardware workingclose to its maximumpotential throughoutthe acquisition phase.The resulting per-formance is well inexcess of the stan-dards yet again withminimal hardwarecosts. This approachdemands much moregeneral purpose pro-cessing capacity butmuch less dedicatedhardware support.

A c q u i s i t i o n

Strategies. We have developed a rangeof strategies to suit GSM and CDMAstandards–based assistance schemes and thehardware architectures previously de-scribed. These involve the signal process-ing schemes described above as well as bit-synchronous signal processing schemes(when bit synchronization is feasible) usingboth correlators and search engines. Weare also working with our chipset part-ners to optimize their hardware architec-tures to suit advanced acquisition strategiesto improve sensitivity and acquisition time.One example is the use of multiple smallersearch engines combined with correlators.This is an ideal compromise when generalpurpose processing capacity is limited.

Memory and Hosting Constraints.In developing the technology, we were re-quired to accommodate a range of mem-ory and hosting constraints. Using the u-Nav chipset, we have implemented anMS-Assisted solution in which the firmware

CIRCLE 20

is serially booted into the on-chip static random accessmemory (SRAM) from the cellbaseband chip. This requiredall of the AGPS firmware torun in 64 Kbytes of programmemory and 60 Kbytes of datamemory. We also have imple-mented MS-Based and assistedconventional solutions utiliz-ing external flash memory forthe program but internal datamemory only.

Utilizing the flexible corre-lator hardware mentioned ear-lier, a firmware solution run-ning on a host processorperforms all of the subATTOprocessing to produce code-phase and Doppler measure-ments or location.

The firmware supportsMS-Assisted, MS-Based, andmulti-mode operation. It runson an ARM 9 microprocessorbut can easily be ported toothers.

It has been uniquely struc-tured to facilitate sharing of the processingresources with foreign, high priority appli-cations as might be found on a cell base-band processor. The firmware adapts grace-fully to the loss of processing capacity andloss of signal samples. It is independent ofthe platform’s real-time operating system(RTOS) and can be collapsed into a sin-gle task.

ConclusionsGPS operation can be enhanced in its per-formance through the provision of assis-tance data wirelessly.

This assistance can be supplied via theuser plane or control plane. It can consistof satellite data gathered from remote GPSreceivers with direct line of sight to thesatellites, coarse receiver position, preciseor coarse time, or more specific data de-rived from these primary elements.

The wireless network may also performthe location solution, although the benefitof this is debatable. The use of assistance canresult in much faster acquisition of weakersignals, and can facilitate navigation solu-

tions that would not otherwise be possible. The benefit of the time assistance is de-

pendent in a complicated way on its uncer-tainty. Both the search strategy and the nav-igation solution have to be adapted to thelevel of this uncertainty.

CDMA inherently facilitates precise timeassistance. GSM networks, however, haveto be augmented to provide it and gener-ally will not be.

On the other hand, CDMA cell sizes canbe much larger than GSM cell sizes, result-ing, in principle, in slower signal acquisi-tion and/or a need for more GPS process-ing capacity. However, this is not asignificant factor in practice.

GSM uses short time slots for transmis-sion and this facilitates interference miti-gation via signal blanking. A CDMA hand-set may employ “antenna switching,” butwith potentially serious impact on bothGPS and voice communication. RF designfor concurrent operation is far preferable.

There are some differences between thestandards developed for CDMA and GSMbut, in general, the assistance supplied and

the performance requirements are quite sim-ilar. However, the sensitivity required by thestandards is well under that required to pro-vide reliable performance under a range ofindoor and urban canyon environments.

Hardware-mechanized search enginesare useful but typically are not flexibleenough to permit the potential of the hard-ware to be fully realized. subATTO signalprocessing allows similar performance tobe achieved with far less dedicated hard-ware support, but requires much more gen-eral purpose processing capacity. A hybridof multiple smaller search engines and cor-relators provide an ideal compromisewhere such processing capacity is not available.

AGPS solutions require a range of dif-ferent hardware and software solutions. Ourteam has experience with hardware employ-ing both correlator fingers and search en-gines, and with flexible correlator hardwaredesigned to work with a host processor. Ourfirmware solutions have ranged from MS-Assisted firmware running in 64 KBytes ofinternal SRAM to taskless, RTOS–inde-

���INNOVATION AGPS

44 GPS World MAY 2005 www.gpsworld.com

¶ Figure 2 A two-code-phase subATTO search engine. In addition to the operational blocks described in Figure 1,the ACV (autoconvolution) block constitutes a proprietary algorithm utilizing multiple fast Fourier transform (FFT)bins. The FFT/ACV/squarer process provides improved sensitivity for the same overall integration period compared tothe technique of Figure 1.

www.gpsworld.com MAY 2005 GPS World 45

pendent multi-mode firmwarerunning on a host processorthat also supports high prior-ity foreign tasks.

AcknowledgmentsThe entire engineering team atSigNav contributed to thiswork through their innovationand tireless dedication. In par-ticular, I wish to thank EamonnGlennon, who has made major algorithmiccontributions. Our colleagues at u-Nav andits partners and customers contributed toour growing collection of insights, as havethose of our other partner who cannot yetbe named. This article is based on the paper“Lessons Learnt in Assisted GPS” presentedat GNSS 2004, the 2004 International Sym-posium on GNSS/GPS, held in Sydney,Australia, December 6-8, 2004. c

BiographyROD BRYANT is the chief executive officer and chieftechnology officer of SigNav Pty. Ltd. in Canberra,Australia. Until 1994, Bryant worked at Auspace Ltd. foreight years in a variety of engineering, management, andbusiness development roles associated with satelliteengineering. He has a Ph.D. in real-time ultrasonic imag-ing from the University of Adelaide.

“Innovation”is a regular columnfeaturing discussions aboutrecent advances in GPS technolo-gy and its applications as well asthe fundamentals of GPS position-ing.The column is coordinated byRICHARD LANGLEY of the

Department of Geodesy and Geomatics Engineering at theUniversity of New Brunswick, who appreciates receivingyour comments and topic suggestions.To contact him,see the “Columnists”section on page 6 of this issue.

¶ Figure 3 Core signal-processing scheme. The fast Fourier transform (FFT) block is equivalent to the ensemble oflocal oscillators, second mixers, and first summers of Figure 2, while the windowing and eliminate data blocks corre-spond to the ACV (autoconvolution) block in Figure 2. Various estimation algorithms are employed depending on theimplementation.

subATTO Tracking SolutionWe have called our solution subATTO because it provides for acqui-sition and tracking of GPS signals weaker than an attowatt (10-18

watts or –180 dBW). This level of sensitivity is essential to providereliable positioning inside buildings, multi-story car parks, and inurban canyons. The sensitivity achieved is highly competitive withrespect to other AGPS schemes reported.

One important aspect of the solution is the subATTO algorithms areintegrated with normal receiver firmware. Under typical conditions,the receiver behaves like a normal, fully autonomous receiver thatacquires the navigation message data in the standard way. ThesubATTO solution takes advantage of this.

In the early 1990s, we developed a patented phase coherent open-loop (PCOL) carrier processing scheme providing faster reacquisi-tion and far greater dynamic resistance than conventional trackingloop techniques. Though considerably refined since the awarding ofa patent in 1991, this scheme was the basis on which we built thesubATTO signal processing solution.

To achieve sensitivity to around –185 dBW (equivalent to a carrier-to-noise density ratio of 19 dB–Hz) or below and be capable of operat-ing in a severe fading environment, it is essential that the signal pro-cessing technique require no prior knowledge of the navigationmessage data. PCOL permits code and carrier tracking completelyindependently of the data extraction, and thus was an ideal basis forthe development of subATTO.

The secret of this approach is to eliminate the data modulation in away that minimizes the loss associated with the non-linear process-ing required.

As illustrated in Figure 3, this is achieved by performing band-limit-ed non-linear operations in the frequency domain. Note that Figure 3is indicative only. Although each output bin is derived from a limitedportion of the input bandwidth, this is actually done in such a waythat the post-correlation bandwidth is preserved. Hence one fre-quency search step typically covers 1 kHz depending on the post-correlation sampling rate. After eliminating the data modulation,the carrier phase and frequency is estimated and these values areused in obtaining the early and late amplitude estimates required forcode tracking.

Refining this scheme to cope with carrier-to-noise density ratios aslow as 19 dB–Hz requires the use of long fast Fourier transforms(FFTs) and the addition of a further layer of complexity to the signal-processing algorithms in the form of various ensembleaveraging techniques.

Another aspect of the subATTO technology relates to the algorithmswe have developed for resolving code phase ambiguity and estimat-ing time without using bit synchronization or the navigation data.The last piece of the jig-saw puzzle is the suite of acquisition strate-gies we have developed to provide optimal trade-offs of time-to-first-fix versus sensitivity under various assistance scenarios using vari-ous hardware facilities.

ADVERTISERS INDEXCompany Page #Atmel inside back coverBEI Systron Donner 29CAST Navigation 23Furuno 6

GNSS/ENC 2005 27GPS City 51Honeywell 21Interstate Electronics Corp. 7ION GNSS ’05 33Javad Navigation Systems

inside front coverLemos International 51

NavCom Technology 11Navtech Seminars & GPS Supply 46NovAtel back coverOmniStar 25Pacific Crest 37Raven 18Rakon 12Rayming 51Septentrio 13Spectrum Control 43Spirent Federal 9u-blox AG 31Verizon 15Waypoint Consulting 20

COMPANY INDEXCompany Page #AETC 34ALAN Electronics 50Astrata Group 47Atmel 47Bones in Motion 39Cambridge Positioning Systems 50Clarity Communication Systems 39Crossbow Technologies 34, 49CSI Wireless 48Dallas Semiconductor 48DAP Technologies 49ESRI 38Fastrax 49Franson Technology 49Furuno 16, 26Garmin 48, 49, 50gate5 38Geo++ 49GPS Innovations 34Heim Data Systems 50Hemisphere GPS 48Hewlett-Packard 39Holux 49Homeland Security Technology 49InfoGation 39iXSea 49Javad Navigation 50

KDDI 38Leica Geosystems 48, 49LOC-AID Technologies 39Maptuit 50Maxim Integrated Products 48Microsoft 38MobinTeleCom 47Motorola 38NAVTEQ 38NEC 39Networks In Motion 39Nextel 38Novariant 47Nova Research 34NovAtel 48, 50NTT DoCoMo 38Outback 48Pacific Crest 34PlanetLink Communications 50Position One Consulting 35Qualcomm 38, 49ReFLEX 49RHS Inc. 48Sarantel 49SigNav 40SiRF Technology 38, 48, 49, 50Smarter Agent 39Socket Communications 38Sony 49Sprint 38Steelwater Solutions 50TCS 39Telcontar 38TeleAtlas 38Thales 47TomTom 49Trace Technologies 49TriaGnoSys 47Trimble 34, 48, 50Tyco Electronics 50u-blox 47, 48u-Nav Microelectronics 43, 45Vivo 38Zeni Lite Buoy 16

46 GPS World MAY 2005 www.gpsworld.com

���INNOVATION AGPS

Further ReadingFor further information on subATTO, see

“A GPS Receiver Optimized for LandVehicles” by R.C. Bryant in Proceedingsof ION GPS-93, the 6th InternationalTechnical Meeting of the SatelliteDivision of The Institute of Navigation,Salt Lake City, Utah, September 22-24,1993, Volume II, pp. 1599-1606.

“subATTO Indoor GPS – Pitfalls,Solutions and Performance Using a

Conventional Correlator” by R. Bryantin GPS Solutions, Vol. 6, No. 3, 2002,pp. 138-148.

For previous GPS World articles on assistedGPS, see

“Wireless-Assisted GPS: Keeping TimeWith Mobiles” by J.T. Syrjärinne in GPSWorld, Vol. 12, No. 1, January 2001, pp. 22-31.

“GPS Reference Networks’ New Role:Providing Continuity and Coverage” by P.

Enge, R. Fan, and A. Tiwari in GPS World,Vol. 12, No. 7, July 2001, pp. 38-45.

“Indoor GPS: The No-chip Challenge” byF. van Diggelen and C. Abraham in GPSWorld, Vol. 12, No. 9, September 2001,pp. 50-58.

“Assisted GPS: A Low-InfrastructureApproach” by J. LaMance, J. DeSalas,and J. Järvinen in GPS World, Vol. 13, No. 3, March 2002, pp. 46-51.

CIRCLE 21

INDEX